Beyond Relevance Ranking: A General Graph Matching Framework for Utility-Oriented Learning to Rank

Learning to rank from logged user feedback, such as clicks or purchases, is a central component of many real-world information systems. Different from human-annotated relevance labels, the user feedback is always noisy and biased. Many existing learning to rank methods infer the underlying relevance of query–item pairs based on different assumptions of examination, and still optimize a relevance based objective. Such methods rely heavily on the correct estimation of examination, which is often difficult to achieve in practice. In this work, we propose a general framework U-rank+ for learning to rank with logged user feedback from the perspective of graph matching. We systematically analyze the biases in user feedback, including examination bias and selection bias. Then, we take both biases into consideration for unbiased utility estimation that directly based on user feedback, instead of relevance. In order to maximize the estimated utility in an efficient manner, we design two different solvers based on Sinkhorn and LambdaLoss for U-rank+ . The former is based on a standard graph matching algorithm, and the latter is inspired by the traditional method of learning to rank. Both of the algorithms have good theoretical properties to optimize the unbiased utility objective while the latter is proved to be empirically more effective and efficient in practice. Our framework U-rank+ can deal with a general utility function and can be used in a widespread of applications including web search, recommendation, and online advertising. Semi-synthetic experiments on three benchmark learning to rank datasets demonstrate the effectiveness of U-rank+ . Furthermore, our proposed framework has been deployed on two different scenarios of a mainstream App store, where the online A/B testing shows that U-rank+ achieves an average improvement of 19.2% on click-through rate and 20.8% improvement on conversion rate in recommendation scenario, and 5.12% on platform revenue in online advertising scenario over the production baselines.

Contributions and Blind Spots of Constructivist Norms Research in International Relations, 1980–2018: A Systematic Evidence and Gap Analysis

The study of international norms from a social constructivist perspective has been one of the major conceptual innovations to the discipline of international relations (IR) over the past forty years. However, despite the concept's ubiquity, there is only a limited understanding of the large-scale trends in research associated with its rise. This analytic essay interrogates conventional wisdom, using a dataset of 7,795 mainstream, English-language journal articles from the Teaching, Research and International Policy Journal Article Database, supplemented with data from Web of Science. How have international norms been studied substantively and methodologically, what are major contributions and blind spots, and which opportunities for future innovation might exist? Although norms research has historically helped expand the scope of issues covered in IR (e.g., gender issues and public health), others have evidence gaps relative to the broader discipline of IR (e.g., terrorism and public opinion). Over the years, the proportion of empirical studies has increased, while purely theoretical, epistemological, and methodological work and innovation have decreased. Despite calls for methodological pluralism, norms research is significantly more qualitative and conceptual than mainstream IR in general and far less multi-method. While more international and less US-based than IR in general, norms research in mainstream journals seems to be no closer to a “Global IR,” measured by regional focus and author affiliation. This suggests three promising avenues for future innovation: greater attention to specific substantive blind spots, more multi-method research, and increased attention to the agenda of Global IR. Beyond these individual insights, this review illustrates the general utility of complementing narrative literature reviews with ones based on quantitative data. It also provides a case study on conceptual proliferation and innovation in IR.

A human fetal lung cell atlas uncovers proximal-distal gradients of differentiation and key regulators of epithelial fates

We present a multiomic cell atlas of human lung development that combines single cell RNA and ATAC sequencing, high throughput spatial transcriptomics and single cell imaging. Coupling single cell methods with spatial analysis has allowed a comprehensive cellular survey of the epithelial, mesenchymal, endothelial and erythrocyte/leukocyte compartments from 5-22 post conception weeks. We identify new cell states in all compartments. These include developmental-specific secretory progenitors that resemble cells in adult fibrotic lungs and a new subtype of neuroendocrine cell related to human small cell lung cancer; observations which strengthen the connections between development and disease/regeneration. Our datasets are available for the community to download and interact with through our web interface (https://fetal-lung.cellgeni.sanger.ac.uk). Finally, to illustrate its general utility, we use our cell atlas to generate predictions about cell-cell signalling and transcription factor hierarchies which we test using organoid models.

Optimal Asset Allocation Subject to Withdrawal Risk and Solvency Constraints

This paper investigates the optimal asset allocation of a financial institution whose customers are free to withdraw their capital-guaranteed financial contracts at any time. In accounting for the asset-liability mismatch risk of the institution, we present a general utility optimization problem in a discrete-time setting and provide a dynamic programming principle for the optimal investment strategies. Furthermore, we consider an explicit context, including liquidity risk, interest rate, and credit intensity fluctuations, and show by numerical results that the optimal strategy improves both the solvency and asset returns of the institution compared to a standard institutional investor’s asset allocation.

On the degree of distortions under second-degree price discrimination

This paper highlights the limitations and applicability of results developed by Chao & Nahata (2015) for nonlinear pricing. Although Chao and Nahata appear to provide necessary and sufficient conditions for general utility functions, we show that one of their results leads only to a restatement of two constraints, and another result may not be valid when consumers can freely dispose of the good. Their model allows for the possibility that higher quantities will have a lower price than smaller quantities. We provide conditions under free disposal that preclude this anomaly. Our analysis suggests that further research on violations of the single-crossing condition should be encouraged.

Estimating disease prevalence in large datasets using genetic risk scores

Clinical classification is essential for estimating disease prevalence but is difficult, often requiring complex investigations. The widespread availability of population level genetic data makes novel genetic stratification techniques a highly attractive alternative. We propose a generalizable mathematical framework for determining disease prevalence within a cohort using genetic risk scores. We compare and evaluate methods based on the means of genetic risk scores’ distributions; the Earth Mover’s Distance between distributions; a linear combination of kernel density estimates of distributions; and an Excess method. We demonstrate the performance of genetic stratification to produce robust prevalence estimates. Specifically, we show that robust estimates of prevalence are still possible even with rarer diseases, smaller cohort sizes and less discriminative genetic risk scores, highlighting the general utility of these approaches. Genetic stratification techniques offer exciting new research tools, enabling unbiased insights into disease prevalence and clinical characteristics unhampered by clinical classification criteria.

Theory and Practice of Quantitative Assessment of System Harmonicity: Case of Road Safety in Russia before and during the COVID-19 Epidemic

People have had an interest in harmony issues for thousands of years; however, there is still no elaborated system of views on these questions. Ancient Greeks understood harmony as an agreement of opposites. A surge of interest in the study of the harmonic aspects of being occurred in the twentieth century due to the development of systems science, particularly regarding synergetic system effects. At the same time, there are still relatively few applications of synergetics because of the absence of an accurate methodology for the identification of system harmonicity. The aim of this research is to develop the methodology for the quantitative assessment of system harmonicity by considering a practical example: the quantitative assessment of the harmonicity of the road safety provision system (RSS) and its dynamics during the last 15 years (2006–2020). In addition, the impact of the COVID restrictions on population mobility in Russia in 2020, on the change in the harmonicity of the road safety provision system, is considered. During the research it was established that the quality factor g of the Russian road safety provision system changed from g2006 = 1.9565 to g2020 = 2.4646, which promoted the decline of the relative entropy of the Russian road safety provision system from Hn RSS 2006 = 0.8623 to Hn RSS 2020 = 0.7553. The deep reason for that change was the modification of relation between ”weights” or the significance of the contribution of different elements of the cause-and-effect chain in the formation of the factual level of the road accident rate in Russia in the last 15 years. The main conclusion of this research is that the harmonicity of the Russian road safety provision system, assessed by the normalized functional general utility GUn, has been increased, and it has already exceeded the level of harmonious reference systems GUn = 0.618. In fact, the normalized functional general utility GUnof the Russian road safety provision system increased from GUn RSS 2006 = 0.615 to GUn RSS 2020 = 0.652 (by 6.0%), from 2006 to 2020. Simultaneously, the share of the normalized used resource Xn declined, allowing a conclusion to be drawn about a significant improvement in the balance ”efficiency-quality” of the Russian road safety provision system. The COVID lockdown played a positive role in this process. Harmonicity of the Russian road safety provision system, assessed by the normalized general utility GUn RSS, increased by 0.46% from 2019 to 2020.

Asymptotic Portfolio Strategy Based on the CEV Model with General Utility Function

The optimal investment problem is a hot field of financial risk control. The analytical solution of investment strategy can be obtained with the power function utility and exponential function utility when the stock price obeys the constant elasticity of variance (CEV) model. However, different investors have different risk preferences; it means that different investors have different utility functions. In this paper, we propose an asymptotic analysis method to obtain the asymptotic solution of investment strategy with the general utility function. The value function is expanded in the form of series, the expressions of the zero-order term and first-order term of the series expansion are derived, respectively, and the error between the asymptotic approximation and the optimal value function is calculated. Finally, the numerical examples provide comparative analysis between the analytical solution and the asymptotic solution to verify the effectiveness of the proposed method.

AlphaFold Protein Structure Database for Sequence-Independent Molecular Replacement

Crystallographic phasing recovers the phase information that is lost during a diffraction experiment. Molecular replacement is a commonly used phasing method for crystal structures in the protein data bank. In one form it uses a protein sequence to search a structure database to find suitable templates for phasing. However, sequence information is not always available, such as when proteins are crystallized with unknown binding partner proteins or when the crystal is of a contaminant. The recent development of AlphaFold published the predicted protein structures for every protein from twenty distinct species. In this work, we tested whether AlphaFold-predicted E. coli protein structures were accurate enough to enable sequence-independent phasing of diffraction data from two crystallization contaminants of unknown sequence. Using each of more than 4000 predicted structures as a search model, robust molecular replacement solutions were obtained, which allowed the identification and structure determination of YncE and YadF. Our results demonstrate the general utility of the AlphaFold-predicted structure database with respect to sequence-independent crystallographic phasing.